EEVblog Electronics Community Forum
Electronics => Metrology => Topic started by: Nuno_pt on September 06, 2016, 01:19:22 pm
-
Hi all,
This is my first try with Metrology, I'm looking at getting an Decade Resistance standard, to check/Cal my DMM till 6.5 digits.
What would be the best way, buy an GR or ESI decade to 100K or 1M, or build one with VHP101/2z till 100K, or maybe Caddock TF020/50N till 10M?
I'm thinking about 100 Ohm, 1K, 10K, 100K, 1M, 10M and 100M as values.
I've look at topics from zlymex, TiN, and some others.
-
With either way you go, you will be dependent on resistance measurement from calibrated instrument.
So you need define what is best for you (cost? accuracy? stability?) and then pick solution which meets needed requirements.
Also there is high chance that vendor calibration of your 6.5d meter be cheaper than getting resistance standard suitable for such verification.
-
Thanks TiN,
Well about the cost of CAL, here they say that the cost would be around 450€ + 23%VAT = 553.50€ for 6.5d meter.
I would think that something in the range of cost vs accuracy vs stability, should be enough.
My first gold was to check if the 2000 is good.
The GR or ESI are around $130 on bay.
I know I still need someone with an Cal DMM above 6.5d to check the Decade, that is the most difficult here since that here most of the people are not involve in Volt-nut, even with Time-nuts only a hand full in the country, so Volt is even worse.
I've a friend that is working on VitrOhm factory here in Portugal (near my house), I know that they have Precision Decades there inside of a control room, Temp & Humi, maybe I should ask him if they have something also for VDC + IDC, and if he could calibrate it there.
Since my 2000 can only be adjust V + I + Ohm.
-
One way is to make simple DIY and have someone calibrated unofficially like these photos.
Edit: I've calibrated(unofficially) many resistance and voltage standards here in China, and one of my 10k standard has traveled for two years among R-nuts.
-
Thanks zlymex,
that was what I was thinking about, making some DIY with the Vishay or Caddock, and see if I can find someone with a 7.5d or maybe an 3458 to read them, and that way I could check my DMM's or maybe CAL them.
The worst still is find some 3457 or 58 in PT, or someone interest in help , to CAL them unofficially.
-
Source for military resistors
https://world.taobao.com/item/15143277363.htm?spm=a312a.7728556.2015080705.20.i4CZ0j&id=15143277363&scm=1007.12006.39884.i39005537941&pvid=472a1c7c-3c9b-450d-b14b-d6b6a4a520c2
-
Uma coisa mais sofisticada
https://world.taobao.com/item/535739458651.htm?spm=a312a.7700714.0.0.xDNaYU#detail
-
Hi
I am in the same/similar predicament as Nuno_pt, I have just got a 6.5 digit meter but is is showing difference between my 5.5 digit meter.
How do I verify which one is in calibration?
I did not think resistance was the way to go, I thought Voltage would be the better thing to concentrat on.
Would I be better off doing the same as Nuno_pt ?
-
Source for military resistors
https://world.taobao.com/item/15143277363.htm?spm=a312a.7728556.2015080705.20.i4CZ0j&id=15143277363&scm=1007.12006.39884.i39005537941&pvid=472a1c7c-3c9b-450d-b14b-d6b6a4a520c2
My first 10V standard(>9 years ago) employ 14 those RX70 resistors for statistical step-up, but proved to be not very good.
Those resistors are made of Manganin with not so small TCR and the drift is not so small too.
-
Uma coisa mais sofisticada
https://world.taobao.com/item/535739458651.htm?spm=a312a.7700714.0.0.xDNaYU#detail
There are cheaper ones such as https://world.tmall.com/item/36662158625.htm?spm=a312a.7700714.0.0.rCrh3H
-
here is a spec sheet for those RX-to series resistors. TC is not very good.
http://www.doublecircle.com/doc_down/200908041016185853.pdf (http://www.doublecircle.com/doc_down/200908041016185853.pdf)
-
here is a spec sheet for those RX-to series resistors. TC is not very good.
Not surprising. Military Spec is for
a) robustness, both physical and electrical
b) wide temperature range (but not necessarily good temperature stability)
They are also not necessarily good precision.
-
I've to find a place to ask for a quote of:
Vishay 101 for 100 Ohm, 1K, 10K, 100K and VHP202Z for 1 Ohm, 10 Ohm.
For 1M, 10M and 100M, I'm thinking about Caddock TF020R/N or TF050R/N.
This will all be put in an Alu box, in 4W with some Pomona banana plugs, with an BME280 sensor inside, and some silica bags.
The Alu box would be put inside of an IP67 Plastic box, this box will have 4W connected to the inside bananas trough some CAT5 cable, solder to the outside bananas and connected to the inside bananas with some copper lugs.
Would this be acceptable?
-
Why the plastic box? That makes all bulky without visible benefits, also adding connection points which have more chance to go bad.
Also if you go 4W up to 100K, that will be 24 connectors just for those six resistors, if you want wire them separately.
for 1M and higher, 2W is ok, but insulation become critical there, so you might want to put those ports on teflon sheet or similar.
Good idea for BME280, easy to interface and small footprint to log data. Perhaps also add EEPROM I2C ROM on same board, so one with Pi can store calibration values/environment data/dates on it too ;).
-
Thanks TiN, I haven't made the math on the connectors before writhe that, now I see the stupidity of that.
So what about 4W till 100K, and separate box for 1M till 100M with 2W.
-
I am planning to build something similar for my home lab (to calibrate my two Keithley 2015 meters). My current idea is to use a single box with a temperature sensor (and possibly even temperature control) for all resistors and connect each resistor to a 9-pin D socket, eight pins for a 4-wire connection (4 pairs) and the remaining pin for the case/ground/guard. This way I can easily select one of the resistors by plugging the cable (9-pin plug to 4+1 banana plugs) in one of the sockets. The whole unit could be quite portable. I have some nice LT450/C wirewound resistors (0.002%, 3ppm/C) that I might use for that purpose.
Cheers
Alex
-
I've send an email to Bedex in Germany asking about the prices of VHP101 and VHP202Z, I'll let you guys know if they give some prices.
-
i was looking at this video here < https://www.youtube.com/watch?v=a6AIoEj5Bzw (https://www.youtube.com/watch?v=a6AIoEj5Bzw) >, and thinking if this would be a good road?
I was looking at Caddock TF or USF370 for 1M, 10M and 5x20M for 100M.
Can you explain why the 9 instead of 10 (e.g. 9 Ohm instead of 10).
My 2000 need to be check at 100 Ohm, 1K, 10K, 100K, 1M, 10M, 100M, and for CAL 1K, 10K, 100K, 1M.
-
Since this is related what about using the Omega OSW series of switches to select the resistor. In their most simple form, they offer switches with 2,3,4 poles. Even when using the 2 pole arrangement you can have the sense/source binding post wired to the switch. That way you only need 4 binding post. These switches are meant for thermocouples/rtd's/thermistors, so they are offered in either gold/silver plated contacts.
-
How do the modern calibrators work for resistance? Surely they aren't as old school cool as EEVBlog #544 - Fluke 5450A?
https://www.youtube.com/watch?v=QFOH8n43kY4&ab_channel=EEVblog (https://www.youtube.com/watch?v=QFOH8n43kY4&ab_channel=EEVblog)
I'm guessing with precise voltage and current sources and a micro you can emulate DC calibration resistors all you want and accurately too?
Of course I have tried messing with my CR on my Maynuo DC load and, yeah not quite precision! :-DD
-
I had three 5450A. I'm planning to tear one down and get rid of those relays, solder those resistors directly to binding posts.
I had another plan to parallel-series connect the 90k string of deck A inside one of my 720A to get a 10k, but proved to be not very stable.
-
Thanks to everyone who gave information. :-+
I do not have access to anything better than the 6.5 digit meter that I have :(
I also do not have enough equipment to build such a stablised, precision 10V ref :-[
I am not too bothered about knowing how imprecise the 5.5 and 6.5 digit meters are at the moment, just (re)learning about metrology.
Oh well :)
-
Thanks to everyone who gave information. :-+
I do not have access to anything better than the 6.5 digit meter that I have :(
I also do not have enough equipment to build such a stablised, precision 10V ref :-[
I am not too bothered about knowing how imprecise the 5.5 and 6.5 digit meters are at the moment, just (re)learning about metrology.
Oh well :)
Where are you in the UK? Perhaps I can lend you one of my 10V references, measured by HP 3458A Opt 002 and stable to about 5ppm for last 6 months or so (including the HP meter drift). It is small and can easily be sent by Special Delivery if you are not local. Drop me a PM if you are interested.
Cheers
Alex
-
Thank you Alex Nikitin, that is a kind offer but I think I will be content with what I have right now.
I know everything is accurate enough to 10mV (thanks to my new Keysight U1232a and UT61E).
Anything beyond that is just a fun learning exercise for me ;)
FYI, I am in Milton Keynes
-
Hi
Some good news (for me at least). I popped in to see the equipment manager where I work and he is allowing me to use the calibrated 6.5 digit meters to verify my AD584. I will then have the confidence in my voltage reference to check against my 5.5 and 6.5 digit meters.
Thanks again to all for the information and offer of help :-+
-
I've to find a place to ask for a quote of:
Vishay 101 for 100 Ohm, 1K, 10K, 100K and VHP202Z for 1 Ohm, 10 Ohm.
For 1M, 10M and 100M, I'm thinking about Caddock TF020R/N or TF050R/N.
........
The smallest value Vishay provides for VHP202Z is 5 Ohm according to the datasheet.
I once try to order 5 Ohm(in the hope to use 5 to make 1 Ohm) and they rejected saying that the actual smallest available was 10 Ohm, so I ordered 20 10-Ohm.
Vishay provides 1 Ohm hermetic Z-foil as VHP4Z and VPR247Z.
AFAIK, Z-foil is better than regular foil only in TCR aspect. However, VHP202Z only guarantee 4ppm/K TCR for 10 Ohm, far more worse than required by a decent resistor standard. In this respect, regular hermetic foil resistors are equally good, since temperature compensation of some kind should be used in order to DIY a good resistance standard anyway.
-
........
You mentioned how you might compensate a negative TCR in another thread [with a copper-wire "resistor", which is what I thought of too], but how do you compensate a positive TCR? NTC thermistors are not very linear-- do you know of a more linear negative compensation material?
I have developed a quick and efficient way to compensate alpha(of both positive and negative) and beta TCR to virtually zero by using hermetic precision grade NTC(and with copper resistor).
Although the value of a NTC is exponential with temperature, the curve is concave, opposite of most of metal-based resistors(such as WW and foil, most of them with negative beta with convex shape). By adding a resistor in series with the NTC and possible another resistor in parallel, I can control the curvature, making the final complementary resistance a 3rd order, with very small resistance variation of less than 0.1ppm in normal lab temperature range(18 to 28 degree C).
Due to the very large variation of NTC with temperature, usually very small 'amount' of compensation is needed, making the NTC and associated resistor very insignificant and easy to choose.
I haven't publish the compensation in English yet(waiting for an actual test result), but available in Chinese(with many charts and tables): http://bbs.38hot.net/forum.php?mod=viewthread&tid=167995 (http://bbs.38hot.net/forum.php?mod=viewthread&tid=167995)
-
The smallest value Vishay provides for VHP202Z is 5 Ohm according to the datasheet.
I once try to order 5 Ohm(in the hope to use 5 to make 1 Ohm) and they rejected saying that the actual smallest available was 10 Ohm, so I ordered 20 10-Ohm.
Vishay provides 1 Ohm hermetic Z-foil as VHP4Z and VPR247Z.
AFAIK, Z-foil is better than regular foil only in TCR aspect. However, VHP202Z only guarantee 4ppm/K TCR for 10 Ohm, far more worse than required by a decent resistor standard. In this respect, regular hermetic foil resistors are equally good, since temperature compensation of some kind should be used in order to DIY a good resistance standard anyway.
Thanks for the heads up zlymex, at the time I must have been looking at other datasheet, and writing this at same time, so the confusion.
You're right it only goes down to 5 Ohm.
I've send a email yesterday to Rhopoint to quote some Caddock USF - 240, 271, and 371 series.
-
Here is my first attempt to get a half-decent 10K reference. I've used this well aged LT450/C (3ppm/C max) resistor and trimmed it to read close to 10K exactly on the HP3458A by adding a 300M shunt (the resistor on its own is +35ppm). So far it looks quite good, with an overnight variation in the lab (not temperature controlled) under 2ppm (including obviously the HP3458A drift).
Cheers
Alex
-
Here is my first attempt to get a half-decent 10K reference. I've used this well aged LT450/C (3ppm/C max) resistor and trimmed it to read close to 10K exactly on the HP3458A by adding a 300M shunt (the resistor on its own is +35ppm). So far it looks quite good, with an overnight variation in the lab (not temperature controlled) under 2ppm (including obviously the HP3458A drift).
Cheers
Alex
Good job!
I got some of those MC wire-wound( and 100M) too that often show upwards drift.
-
I got some of those MC wire-wound( and 100M) too that often show upwards drift.
Yes, wire-wound resistors do apparently drift up in time. However in this case it is about +70ppm (maximum, taking in the uncertainty of the meter, and could be below +15ppm ) drift over 27 years, or +2.5ppm/year . I can live with that .
I've bought a set of these resistors, new old stock, from 1 Ohm to 1M (x1,x2,x4,x8 values for six decades). I will make the rest of the decades with these - 1, 10, 100 Ohm, 1K (these with a 4-way connection), 100K and 1M. I also have 10M 0.01% 5ppm Caddock as well as a couple of 1% 1G and 100G resistors, so only 100M and 10G are missing at the moment for a full 12 decades range.
Cheers
Alex
-
Here is my first attempt to get a half-decent 10K reference. I've used this well aged LT450/C (3ppm/C max) resistor ...
Why do you think it's 3ppm/C max ? I has tons of Mann Components PWW resistors (LT, UP, DUP etc) and have never met the official datasheets. The same problem with the Kelvin PWW resistors, widely used in Datron equipment.
-
Here is my first attempt to get a half-decent 10K reference. I've used this well aged LT450/C (3ppm/C max) resistor ...
Why do you think it's 3ppm/C max ? I has tons of Mann Components PWW resistors (LT, UP, DUP etc) and have never met the official datasheets. The same problem with the Kelvin PWW resistors, widely used in Datron equipment.
That is what the datasheet says. From my initial measurements it looks like it is better than 1ppm/C around the 22-25C range. I will measure the R(t) curve soon and will know for sure.
Cheers
Alex
-
That is what the datasheet says. From my initial measurements it looks like it is better than 1ppm/C around the 22-25C range. I will measure the R(t) curve soon and will know for sure.
All of my measurements says the same - TC is less than 1 ppm/C. But can you give a link to a datasheet?
-
That is what the datasheet says. From my initial measurements it looks like it is better than 1ppm/C around the 22-25C range. I will measure the R(t) curve soon and will know for sure.
All of my measurements says the same - TC is less than 1 ppm/C. But can you give a link to a datasheet?
Here it is. (http://www.datasheets360.com/pdf/2877094002159733120) You should bear in mind that the tempco is defined for the full temperature range, which is up to 135C for these resistors, and should be considerately lower around 25C.
Cheers
Alex
-
For a reference the TC is not that important. It should be reasonably low, but temperature measurement / stabilization is not unreasonable. If really good there are also influences from humidity and maybe magnetic fields that get important.
The more important factor is long term stability.
-
It is also my conclusion that long-time stability and humidity effects are more critical than temperature sensitivity.
Has anyone some figures to share for Wire Wounds? I have only tested General Resistance 8E16 and Ultrohm and in both cases I have been disappointed by humidity sensitivity. For the old 8E16 I have, long-term stability is far better than the humidity sensitivity. For the Ultrohm I can´t determine the long-term drift as the humidity sensitivity is so high and I bought them last year.
Lars
-
Long term stability is the most important factor in my regard too.
Humidity is an very annoying factor for non-hermetic resistors that it not only has delayed effect but I suspect there are long term effect too.
When I DIYed an 10k standard resistor by using 8 Fluke 19.999k wire-wound 9 years ago, It was very difficult for me to characterize it for some years. I gave up it and decided to use hermetic resistors since for all my later DIY on less than 100 Meg resistance standard.
-
Long term stability is the most important factor in my regard too.
Humidity is an very annoying factor for non-hermetic resistors that it not only has delayed effect but I suspect there are long term effect too.
Why not solder up some hermetic enclosures then? Obviously fine trim would have to be external, but you could seal the main resistors in for good.
-
Long term stability is the most important factor in my regard too.
Humidity is an very annoying factor for non-hermetic resistors that it not only has delayed effect but I suspect there are long term effect too.
Why not solder up some hermetic enclosures then? Obviously fine trim would have to be external, but you could seal the main resistors in for good.
I tried, but I cannot find a good and true hermetic enclosure.
-
Tin plate with soldered seams and ceramic feedthrough caps or glass feedthroughs?
Edit: even thick copper or brass foil (or thin sheet) at a pinch
-
Tin plate with soldered seams and ceramic feedthrough caps or glass feedthroughs?
Edit: even thick copper or brass foil (or thin sheet) at a pinch
Let's say I was lazy. I once teardown an tin and ceramic sealed cap and found out it was not a true hermetic seal(see photo).
I did buy some big oil filled caps(like the 2nd photo) that I may use for sealing 100 Meg+ resistors.
I have already got all the hermetic resistor elements(decade only, with redundancy) necessary to build standard resistors from 0.1 Ohm to 10 Meg.
-
Well you could do, but it looks a lot messier than soldering something up from thin sheet. After all, all a hermetic resistor is is a standard resistor in a solder sealed can with glass to metal seals.
I think solder-in ceramic feedthrough caps are hermetic and someone found a source of glass to metal seals on ebay a while back.
-
Well you could do, but it looks a lot messier than soldering something up from thin sheet. After all, all a hermetic resistor is is a standard resistor in a solder sealed can with glass to metal seals.
I think solder-in ceramic feedthrough caps are hermetic and someone found a source of glass to metal seals on ebay a while back.
True, I always looking for containers with two glass to metal seals. Another thing, it must have a large cap sealed by tin in order for the content to put inside in the first place.
-
I can certainly relate to the 'laziness factor' of having something like that already available. I was just thinking of the effort of cleaning the sticky mess out. :)
I guess there are two other possible factors that might be relevant:
1. A thicker can might be more resistant (but certainly not immune!) to the effects of varying atmospheric pressure, assuming this might be a possible factor affecting drift.
2. The thicker the walls, the greater the internal temperature rise during the final sealing. It would be nice to be able to solder this very quickly to avoid drifting the resistors.
Just random thoughts anyway.
Edit: Ah, found some seals: http://www.ebay.co.uk/itm/QTY-25-HSC-HERMETIC-SEAL-CORPORATION-2709-4-EP40-1-TERMINAL-BOARD-04820-/152026811217?hash=item2365810b51:g:rMoAAOSwPhdVFzYr (http://www.ebay.co.uk/itm/QTY-25-HSC-HERMETIC-SEAL-CORPORATION-2709-4-EP40-1-TERMINAL-BOARD-04820-/152026811217?hash=item2365810b51:g:rMoAAOSwPhdVFzYr)
-
I can certainly relate to the 'laziness factor' of having something like that already available. I was just thinking of the effort of cleaning the sticky mess out. :)
I guess there are two other possible factors that might be relevant:
1. A thicker can might be more resistant (but certainly not immune!) to the effects of varying atmospheric pressure, assuming this might be a possible factor affecting drift.
2. The thicker the walls, the greater the internal temperature rise during the final sealing. It would be nice to be able to solder this very quickly to avoid drifting the resistors.
Just random thoughts anyway.
Edit: Ah, found some seals: http://www.ebay.co.uk/itm/QTY-25-HSC-HERMETIC-SEAL-CORPORATION-2709-4-EP40-1-TERMINAL-BOARD-04820-/152026811217?hash=item2365810b51:g:rMoAAOSwPhdVFzYr (http://www.ebay.co.uk/itm/QTY-25-HSC-HERMETIC-SEAL-CORPORATION-2709-4-EP40-1-TERMINAL-BOARD-04820-/152026811217?hash=item2365810b51:g:rMoAAOSwPhdVFzYr)
Agreed with the 2nd completely. For the 1st, if there is vacuum inside, there would be no pressure effect. Most of hermetic are filled with oil or inert gas that will transfer the atmospheric pressure inside, and the thicker the wall the better. However, modern wires is not suffer from pressure effect very much. The old double-wall standard do suffer main owning to the fact that the wire is tightly wrapped around the inner wall which deform with pressure change.
As for the ebay seals, I don't think they are very good because:
- either the conductor is Kovar, which has very large thermal EMF,
- or the conductor is copper, which gradually build up small cracks because the difference in thermal expansion.
The correct way to build a good glass to metal sealing is to use Kovar tubes, allowing copper wire go thru, thus solve both problems.
-
Those on ebay are Kovar, I got some of those while ago.
-
These are probably what you want - and they're cheaper:
http://www.ebay.co.uk/itm/Russian-military-miniature-glass-Feedthrough-insulators-lot-of-100-/251041737116?hash=item3a7340e99c (http://www.ebay.co.uk/itm/Russian-military-miniature-glass-Feedthrough-insulators-lot-of-100-/251041737116?hash=item3a7340e99c)
-
Ah yes! Those were the ones that came up in that previous thread (someone trying to do something with a vacuum chamber iirc) they allow you to run copper right the way to the resistors. :-+
Thanks splin, they didn't come up in my ebay search. Damn I'll probably have to order some now... just in case. :palm:
-
Ah yes! Those were the ones that came up in that previous thread (someone trying to do something with a vacuum chamber iirc) they allow you to run copper right the way to the resistors. :-+
Thanks splin, they didn't come up in my ebay search. Damn I'll probably have to order some now... just in case. :palm:
This was the thread - it was me wanting to hermetically seal some resistors:
https://www.eevblog.com/forum/projects/low-cost-hermetic-seals/ (https://www.eevblog.com/forum/projects/low-cost-hermetic-seals/)
Naturally, having bought those seals I haven't got round to using them yet due to not being able to make up my mind as to exactly which values and configurations of resistors to can. There are some interesting replies, especially those about epoxy seals, so worth a quick scan for those with an interest in the subject.
-
Some more information on epoxy seals with the interesting statement (wrt helium leak testing):
Most high-performance glass or ceramic seals have leak rates of less than 1x10-5 std cc/sec at 1 atmosphere differential. Pave epoxy seals have leak rates typically between the 10-7 to 10-10 std cc/sec. This means that the leak rates of epoxy seals as measured by the helium test are as low or lower than those of glass or ceramic seals.
Of course 'glass or ceramic seals have leak rates of less than 1x10-5 std cc/sec' doesn't mean much - it could mean that most glass or ceramic seals have leak rates well below 10E-10. I'm not sure I believe that 1E-5 figure, given that leaks > 5 x 10-6 std ccs/sec are classed as 'gross leaks'.
http://www.pavetechnologyco.com/pdf/hermetic.pdf (http://www.pavetechnologyco.com/pdf/hermetic.pdf)
From this:
http://nepp.nasa.gov/workshops/etw2013/talks/Tue_June11_2013/1630_McManus_Pressnell_Hermeticity%20Leak%20Testing.pdf (http://nepp.nasa.gov/workshops/etw2013/talks/Tue_June11_2013/1630_McManus_Pressnell_Hermeticity%20Leak%20Testing.pdf)
If I'm interpreting this correctly, a 1cc enclosure with an 'equivalent standard leak rate' of 1E-7 atm cc/s air will exchange 50% of its atmosphere within 80 days which isn't very long. I would have expected a hermetic seal to be much better than that. In any case it seems that a 1E-8 seal and larger enclosures, say > 10cc, are required for lifetimes of more than a few years.
-
Aren't those specification with vacuum on one side...?
If box is pressurised on the inside, seals would have to hold only differential pressures, making leakage miniscule...
I apologise it if my question is stupid, I find this topic interesting but haven't got much knowledge about it.. :-//
-
pressure isn't pressure
If there is hydrogen on one side and no hydrogen on the other side for example. There will be diffusion no matter of the partial pressure of other gases.
-
The leak rates (especially when small) also depends on the gas. This is why they often test with helium. The leakage rates for water vapor are usually much smaller than for helium, but difficult to measure.
The leakage rates are usually for vacuum and with less pressure difference, leakage can also go down. Though there is still the partial pressure problem, so still some tendency for humidity to come into a dry container, even if pressurized inside. A sealed container will also see pressure changes, from atmospheric pressure and temperature changes.
For a resistor circuit it is not so much keeping any humidity out, but to hold it low at a value below about 40% RH - the low value because often surface leakage gets significant better below a limit of about 50% (depends on the material).
For the volume, a small amount of getter material can absorb quite a lot of humidity, so increasing the effective volume to really large values. This also makes is reasonable to start dry.
Form my experience with vacuum, if would expect epoxy sealing to be good enough.
-
I understand a physics behind it, but have no material knowledge...
So you're saying NASA tested container that had same pressure in and out, just helium inside and then tested how much of it diffused...
That would be a good measure of material porosity, yes...
But here we would be interested in water vapour diffusion rate, which should be slower ..
Also air has different water vapour capacity at different temperatures and air pressure...
Would desiccant in the box be able to help (together with epoxy seals) or it will deteriorate long term (it will saturate eventually?)?
-
I wonder if it would be better to introduce a non-oxidising atmosphere, such as dry Nitrogen or even better, an inert gas like Argon. I suppose it's possible that oxidation could play a part in long term drift of non-hermetic parts, in addition to the shorter term drifts caused by humidity changes.
Argon is readily available for TIG welding as well as some food (wine) preservative products. It is also significantly denser than air, so even less likely to diffuse through seals. It doesn't sound too difficult to introduce two tiny holes in the container so that it could be flushed through and quickly sealed with a couple of solder blobs.
-
as it is a noble gas [doesn't chemically react with most things]
In fact, nothing so far established at temperatures warmer than 40k, so you've got a couple of hundred degrees to spare. ;D
Edit: Argon is 37% denser than air but has much lower thermal conductivity... I have no idea whether that would be better or worse for convection currents (thinking heated Vref type applications here).
-
A desiccant inside can help a lot, as it can adsorb really large quantities of water (e.g. gramms for a small bag). A small bag of desiccant could thus be equivalent to a volume of about 0.1 m3, when it comes to absorbing water. They are long time stable at normal temperatures. There could be a slight problem if the unit gets warm, as this will shift the humidity to higher values, or would need a really dry bag and thus less possible adsorbtion.
There would also be the option to use dehydrated salts / boron oxide instead - can be even more efficient at low humidity levels. It could be just difficult to keep dust away.
There is a more or less smooth transition from small leaks to material porosity. Some of the leakage is just diffusion through bulk material, especially with something like epoxy or as a really bad material silicone. Transport through microscopic cracks is similar and may include short distance through the material. The standard for testing is helium, because it is usually the fastest and there is a low background. Water is rather difficult to measure because there is so much background and it sticks to surfaces - so if you have a small vial, more water could be at the surface than in the volume gas. The adsorbed surface layer of water also influences surface leakage.
For the DIY glass sealed resistors, I would initially keep the other end open. This allows testing for leakage and than filling with the right gas and sealing the glass end as a last step. This is relatively easy by pulling it thin and twisting. This also avoids gas leakage, no matter which orientation as you have the tube end to fill.
-
Well today I got a quote from Rohpoint for the Caddock resistors.
USF240-1.00M-0.01% - £33.11
USF240-10.0M-0.01% - £33.11
USF371-25.0M-0.01% - £42.85 x 4 = £171.40
All of them have the delivery schedule for 12 weeks from now.
-
Well today I got a quote from Rohpoint for the Caddock resistors.
USF240-1.00M-0.01% - £33.11
USF240-10.0M-0.01% - £33.11
USF371-25.0M-0.01% - £42.85 x 4 = £171.40
All of them have the delivery schedule for 12 weeks from now.
Digikey has USF340 in stock. If you can measure the actual value, 0.1% version is about half of the price for 0.01% .
Cheers
Alex
-
I ask the price of this because of the TC ~2ppm/ºC, to see if was much difference in price.
Since the price is the double, and all the resistors will be equal, 4x25M to make 100M, and the 1 & 10M, the .1% 5ppm/ºC vs .01% 2ppm/ºC shouldn't present a problem, correct ?
-
If you want a 100Meg resistor with best TC possible you can ask Caddock to match some USF370 for you.
I bought some matched sets of 5 USF370 with 25Meg, 0,01%, <0,003ppm/V and guaranteed below 1ppm/K. One of these sets cost about 250€+VAT. Not much more than 5 USF370 alone. Maybe you should ask Caddock for a quote.
-
Thanks e61_phil, that price is not more the what they are asking for the 4x25Meg, since they are asking £171.40+VAT.
I've send an email to sales@caddock but the email bounce, and was not deliver.
Still waiting for a replay from VPG with the quote.
-
Just in case anyone's interested, I've just spotted some nice cheap and compact glass hermetic feedthroughs from a seller in Poland. They're much less bulky than the previous ex-soviet ones that have been discussed. Outer diameter is a mere 5mm, with solderable center hole of 0.8mm. Being hollow there's no worry about thermocouple effects. They come in lots of 10 but no additional postage for multiples. Looking at other items, he also has some bigger ones. There are a couple of other sellers in Poland too, someone must have unearthed a crate.
http://www.ebay.co.uk/itm/10pcs-Glass-Hermetic-1-Pin-Feedthrough-Feedthru-Military-Product-/322318095449?hash=item4b0ba7d459:g:X-YAAOxy4t1SoC-4 (http://www.ebay.co.uk/itm/10pcs-Glass-Hermetic-1-Pin-Feedthrough-Feedthru-Military-Product-/322318095449?hash=item4b0ba7d459:g:X-YAAOxy4t1SoC-4)
Now, do I ask Santa for some Argon? :D
P.S. No connection with the seller.